Optimization of a solar chimney design to enhance natural ventilation in a multi-storey

office building

M. Gontikaki BSc. dr. Dipl.-Ing. M. Trcka prof.dr.ir. J.L.M. Hensen Ir. Pieter-Jan Hoes

Unit Building Physics and Systems

Eindhoven University of Technology

Eindhoven, Netherlands

ABSTRACT

Natural ventilation of buildings can be achieved with solar-driven, buoyancy-induced airflow through a solar

chimney channel. Research on solar chimneys has covered a wide range of topics, yet study of the integration in

multi-storey buildings has been performed in few numerical studies, where steady-state conditions were

assumed. In practice, if the solar chimney is to be used in an actual building, dynamic performance simulations

would be required for the specific building design and climate. This study explores the applicability of a solar

chimney in a prototype multi-storey office building in the Netherlands. Sensitivity analysis and optimization of

the design will be performed via dynamic performance simulations in ESP-r. The robustness of the optimized

design will be tested at the final stage, against e.g. windows’ opening by users. This is an ongoing project;

calibration of the solar chimney model and preliminary sensitivity analysis results are presented here.

INTRODUCTION

The exploitation of sustainable energy sources to

cover the functional demands of buildings (for

heating, ventilation, cooling etc) can contribute to

significant energy savings and thus to alleviation of

the current environmental, economical and social

problems related to conventional energy practices.

Passive (natural) ventilation of buildings is a

successful means to save energy otherwise consumed

for mechanical ventilation and/or cooling. Solar

chimneys (SC) are passive elements that make use of

the solar energy to induce buoyancy-driven airflow

and naturally ventilate the building.

A SC differs to a conventional chimney in that at

least one wall is made transparent; solar radiation

enters the chimney through the glazed part and heats

up the walls. The temperature of the air inside the SC

channel rises due to heat transfer from the walls and

the resulting buoyancy drives the airflow through the

channel. The SC pulls air from the interior of the

building, which is replaced by fresh air through

openings or other paths, and natural ventilation is

accomplished. Performance of the SC is primarily

described by the induced ventilation flow rates; in

case heat harvesting is also of interest, air

temperature in the channel is the other important

performance indicator.

The geometry of the SC channel is described by

its height, length and cavity width. The elements of

the SC are presented in Fig. 1; the terms indicated in

Fig. 1 will be used in the following sections to refer

to the solar chimney’s parts and geometrical features.

Integration of a solar chimney in a building is

possible in many ways (Fig. 2), e.g. as part of the

south-facing façade (or the façade where maximum

solar availability applies), on the roof (also known as

‘roof solar chimney’), in the place of a conventional

chimney or as extension of a double façade (usually

for multi-storey buildings).

The present study explores the passive

ventilation potential of a SC integrated in a multi-

storey office building in the Netherlands. It forms

part of a broader project under the name ‘Earth, Wind

& Fire’, concerning the exploitation of solar,

geothermal and wind energy in a prototype building.

This is an ongoing research and results are not yet

available. The following sections present an

extensive literature review, the applied methodology

and envisaged results.

Figure 1. Cross-section and top view of a SC.

Figure 2. Integration possibilities of SC in buildings.

LITERATURE REVIEW

General Literature Review

During the past few decades numerous research

studies have been performed on the SC concept, that

have deepened our knowledge on the performance,

the parameters that affect the performance and the

applicability of the system for natural ventilation and

passive cooling of spaces. The studies can fall under

the broad categories of experimental, analytical and

numerical modeling studies. A general overview of

previous studies based on the above categorization

will be presented initially. This section offers a

general impression of the various research topics and

approaches, and no findings will be reported for

brevity reasons. Instead, the findings of parametric

analysis studies and of the ones regarding integration

in multi-storey buildings will be grouped and

presented in two consecutive separate sections, since

they are greatly related to the present study.

Experimental studies.

Field studies concerning the ventilation induced

by SC configurations in a small-scale single-room

house were performed in Thailand (Khedari et al.,

2000). In a consecutive study Khedari et al. (2003)

studied the performance of SC systems in an air-

conditioned small-scale house. Barozzi et al. (1992)

tested a 1:12 small-scale model of a building where

the roof functions as a solar chimney. Hirunlabh et al.

(1999) performed a similar study for a 31m3 volume

house where a metallic solar wall was placed on the

southern façade. Laboratory experiments on small-

scale models were also performed, where temperature

distribution of the air and the walls’ surfaces, the

velocity profile in the chimney’s channel and the

magnitude of the air flow rates were studied, in

combination with parametric analysis (Chen et al.

2003, Burek and Habeb 2007). The performance of a

full-scale 2m high SC that was part of a 12m3 room,

was studied under laboratory conditions by Bouchair

(1994), while Arce et al. (2009) studied a 4.5m high

model, under real meteorological conditions in Spain.

Analytical studies.

In a number of analytical studies mathematical

models were developed for the prediction of the solar

chimneys’ performance. Experiments were usually

performed, in order to validate the models. Bansal et

al. (1993) developed a steady-state mathematical

model to calculate the mean air flow rates and

temperatures of a solar chimney connected to a

conventional one. Wall temperature was assumed

uniform. Awbi and Gan (1992) analytically derived

the air temperature distribution along the SC channel

by making the same assumption. A steady-state

mathematical model was also developed by

Bassiouny and Koura (2008) to predict the air

changes per hour1 (ACH) induced in a room by a SC.

The thermal resistance network approach was

adopted to set up the steady-state heat transfer

equations in the model of Ong and Chow (2003),

which predicts the performance of a solar chimney

under varying ambient conditions. Mathur et al.

(2006) followed a similar approach to calculate the

ACH induced in a room by a small-sized solar

chimney integrated to a normal window. A simplified

thermal model and a computer program to calculate

ventilation flow rate through a chimney were

developed by Afonso and Oliveira (2000) which also

takes into account varying climatic conditions, the

thermal inertia of the back wall and dependence of

the convective heat transfer coefficient on the air

flow rate.

Numerical modeling studies.

The buoyancy-driven airflow and heat transfer in

vertical heated cavities were investigated

numerically, mainly through Computational Fluid

Dynamics (CFD) simulations in an increasing

number of studies during the past decade. Some of

the studies focused on the airflow regime and the heat

transfer mechanisms in the SC channel, while others

on parametric investigations. The latter will be

mentioned here briefly, since more details will be

given in the following section. Numerical

investigations on the airflow within the Trombe wall

channel, which can be considered a type of SC,

preceded by many years, with the first one being that

of Borgers and Akbari (1979). Rodrigues et al.

(2000) studied the airflow regime and the transition

from laminar to turbulent flow in a SC channel using

a finite-volume method. The heat transfer

mechanisms (by conduction, convection and

radiation) in the SC channel were studied in detail,

using the numerical method of finite-difference

control volumes, by Nouanégué and Bilgen (2009).

In certain studies numerical modeling was

employed to perform parametric analysis on the SC

system. The effects of solar heat gain and glazing

type (Gan and Riffat, 1998), cavity width (Gan,

2006), inclination and low emissivity coating (Harris

and Helwig, 2007) were some of the issues

investigated using CFD modeling. Recently Gan

(2010) found that the size of the computational

domain (for CFD simulations) has an impact on the

airflow and heat transfer coefficient predictions in

solar heated cavities, and made corresponding

suggestions. An extensive parametric analysis for a

SC integrated in a prototype residential building was

performed by Ho Lee and Strand (2009), using

numerical modeling in the program EnergyPlus. The

authors developed a SC module for the program and

1 Ventilation rate expressed as a factor of the space volume:

ACH [-] = (ventilation induced in one hour –[m3])/(space

volume – [m3])

performed single-day dynamic simulations, assuming

three different locations for the building. Numerical

investigations have also been performed to study the

application of solar chimneys or similar

configurations of solar heated cavities (e.g. double

façades) in multi-storey buildings (Punyasompun et

al. 2009, Ding et al. 2005, Letan et al. 2003, Gan

2006). The corresponding findings of these and of the

parametric analysis studies are reported in the

following two sections.

Parametric Analysis Findings

In parametric analysis, parameters are varied one

by one to determine the sensitivity of the system’s

performance against each. It has been employed in

the vast majority of studies and has played a

substantial role in understanding the mechanisms

controlling the performance of the solar chimney and

their interactions. The most influential (and thus

mostly studied) parameters can fall under the

category of geometrical (e.g. wall height, gap width),

construction (e.g. type of glass, insulation) and

climatic (e.g. solar radiation, wind) parameters. Some

indicative results will be presented in this section,

regarding the effect of these parameters on the

system’s performance.

Effect of wall height and cavity width.

The effect of wall height and cavity width has in

some cases been studied and expressed separately,

while in others inextricably, in terms of the height-to-

width aspect ratio (also referred to as height-to-gap

ratio). Air flow increases with height, since the back

wall’s heat gains increase: in a parametric study on

Trombe walls, Gan (1998) found that an increase in

wall height by a quarter is equivalent to an increase

in heat gains by three quarters. In the study of Ho Lee

and Strand (2009) air flow rates increased in all three

assumed locations of the building, by ~73% when the

wall height increased from 3.5 to 9.5m (a cavity

width of 0.3m was considered).

With respect to the cavity width, it was found

that the induced flow rate increases with increasing

width. In the analytical study of Ong and Chow

(2003) it was estimated that a 0.3m wide channel

induced 56% higher flow rate than one of 0.1m.

Similarly, according to the model of Bassiouny and

Koura (2008) a threefold increase in cavity width

causes an increase of ACH by 25%. Rodrigues et al.

(2000) argued that airflow rates grow with cavity

width, but the growth ratios decrease as the width

increases. In some studies an optimum cavity width

(or optimum height-to-width ratio) was reported; for

this optimum width the flow rate became maximum,

while for wider gaps reverse flow occurred that

reduced the mean flow through the SC (Spencer et al.

2000, Chen and Li 2001). This discrepancy between

studies is attributed to the different ranges of height-

to-width ratios investigated and to the influence of

the inlet size. The term ‘inlet’ is used here to describe

the opening from where the room air is admitted into

the SC channel. In case the inlet size increases along

with the cavity width, a lower pressure drop in the

inlet can counterbalance the reduction in the flow rate

caused by reverse flow, so that no optimum width is

found (Gan 1998, Chen et al. 2003). Gan (2006) later

reported an optimum width of 0.55m for a 6m high

SC and supported that the optimum width increases

with height.

Effect of inlet size.

The effect of inlet size was also explored in

some studies, and was found to have a wicker effect

on the performance compared to that of the cavity

width. Increasing the inlet size by a factor of three

increased the induced ACH by 11% in the analytical

study of Bassiouny and Koura (2008), while

Nouanegue and Bilgen (2009) found that volume

airflow rate is a decreasing function of inlet size. Gan

(1998) claimed rising airflow rates for cavity width

beyond 0.3m, as long as the inlet size has the same

size as the width.

Effect of inclination angle.

Various inclination angles (15o/30

o/45

o/60

o) were

studied experimentally by Chen et al. (2003) for a

constant height, cavity width and uniform heat flux.

The air velocity profile across the cavity width was

found to be more uniform when the SC was inclined,

leading to lower pressure losses at inlet and outlet,

and thus to higher airflow rates (45% higher airflow

rate was found for the angle of 45o). Harris and

Helwig (2007) numerically studied the consequences

of inclining the SC along the roof line of buildings

(for the latitude of Edinburgh, Scotland). They

argued that although the heat gains can be favored

due to tilt, heat transfer between air and glazing is

higher, resulting in higher heat losses that could

reduce the performance. Higher flow rates by 11%

were found (for the optimum angle of 67.5o), while

the performance at 45o angle was almost the same as

that of the vertical SC. It is implicit that the impact of

SC inclination is highly dependent on the latitude of

the location.

Effect of back wall properties.

Increasing the thickness of the back wall (i.e its

thermal mass) favors night-time ventilation because

the time distribution of the induced airflow rates

changes; their magnitude on the other hand is not

significantly altered (Charvat et al. 2004, Martí-

Herrero and Heras-Celemin 2007). Insulation of the

back (and side) wall(s) is imperative to prevent

devastating heat losses of the SC (as well as

overheating of the adjacent spaces): airflow rates

decreased by 60% without back wall insulation

according to the study of Afonso and Oliveira (2000),

while Gan (1998) reported that 40% of available heat

gains would be lost through a non-insulated back

wall (0.3m thick). In that study the surface

temperature of the back wall rose by 9oC when

insulation was applied. Back wall absorptivity should

be as high as possible to maximize solar heat gains.

Increasing the absorptivity from 0.25 to 1, led to

increased airflow rates by 42-57% in all three

locations assumed for the building (Ho Lee and

Strand 2009). Back wall emissivity should be as low

as possible to restrict radiative heat losses; when a

low-e coating was applied at the back wall, airflow

rates increased by 10% in the study of Harris and

Helwig (2007).

Effect of glazing type.

The use of double glazing can prevent

downdraught (and thus reverse flow) along the cold

glass surface during winter, but was also found to

increase the induced airflow rates up to 17% when

applied to a Trombe wall used for passive ventilation

in the summer (Gan, 1998). Harris and Helwig

(2007) argued that double glazing improves the

performance of a SC but only marginally, so that it is

not a cost-effective measure (only summer conditions

were considered).

Effect of climate: solar intensity and wind.

The intensity of solar heat flux is the motive

force for the operation of the SC and is thus the most

determinant factor for its performance. In the

experimental study of Chen et al. (2003) varying

values were considered for the uniform heat flux on

the back wall and the airflow rate was found to rise

by ~38% for a threefold increase of heat flux (from

200W/m2 to 600W/m

2). Mathur et al. (2006) found

that airflow rate increases linearly with solar

radiation and Bansal (1993) estimated that a SC with

surface area of 2.25m2, would induce 100m

3/hr and

350m3/hr for solar radiation of 100W/m

2 and

1000W/m2 respectively. Based on experimental

investigations on a small-scale SC, Burek and Habeb

(2007) derived that the mass flow rate is proportional

to Qi0.572

where Qi (W/m2) the uniform heat flux

supplied to the back wall. The simulations of Ho Lee

and Strand (2009) showed flow rates to vary as much

as 200% between the three assumed locations of the

building, as a result of the varying solar availability.

Wind is the second most influential climatic

parameter, as it can create positive or negative

pressures at the outlet of the SC and thus obstruct or

enhance the airflow. In the outdoor experiments by

Arce (2009) at a full-scale SC, the highest airflow

rates coincided with the highest recorded wind

velocity, while Afonso and Oliveira (2000) altered

their model to include wind effects which proved

significant (error between model predictions and

measurements was then lower than 10%). In the

analytical study of Mathur et al. (2006) the large

error of 23% was attributed to wind effects, which

were neglected in the model. In practice, devices can

be incorporated at the outlet of the SC so that wind of

all directions creates negative pressures.

Integration in Multi-Storey Buildings

Investigating the performance of a SC in a multi-

storey building holds great interest due to the high

potential benefits involved in actual, large-scale

applications. This subject has been studied only

numerically in the past few years, mainly with CFD

simulations assuming steady-state temperature and

air velocity fields. The CFD models were validated

against small-scale experimental data.

Passive ventilation of a five-storey building was

found feasible in the study by Letan et al. (2003).

Numerical simulations were performed with

commercial CFD software assuming steady-state

temperature and velocity fields and summer

conditions. It was found that the induced airflow rates

decreased with floor height: the lower the floor, the

higher the airflow rate. Acceptable thermal conditions

were found for the first four floors with air

temperatures close to that of the incoming air (at

maximum 25.5oC), while the fifth floor was

overheated up to 29.9oC. The authors argued that

extending the SC by one storey height would improve

the situation, but mechanical ventilation would still be

required for the top floor.

Ding et al. (2005) investigated a prototype eight-

storey office building with an atrium on the northern

side and a double-skin façade on the southern side,

which extends to a SC channel above the roof height.

Reduced scale experiments and CFD simulations

were employed in this study. The size of the openings

connecting the atrium to the occupant spaces were

fixed, while for the size of the openings between the

floor spaces and the double-skin channel three values

were tested (1m2/2m

2/4m

2 per floor). Different

heights for the SC were also examined. Here as well

the airflow rates got lower with floor level. By

increasing the openings’ size higher airflow rates

resulted at all floors, but the pressure difference at the

upper floors got lower, resulting in lower percentage

of the overall flow going through these. The authors

recommended that the SC should be at least two-

storey high.

Ventilation of a four-storey building (total height

of 12m) via buoyancy-induced flow in a double-skin

façade channel was studied by Gan (2006) using

CFD analysis. Although not a SC channel was

studied here, the double-skin façade works in a

similar way, with the difference of more symmetrical

heating of the two glass skins. Varying cavity widths

for the double-façade channel and different exhaust

configurations from the floors to the channel were

investigated. Replacing the outer skin of the façade

with photovoltaic panels (PV) was also examined.

For the design where each floor has its own opening

Figure 3. Applied Methodology.

to the double-skin channel (referred to as ‘room

ventilation’ in the study), flow rates decreased from

bottom to top, due to lower stack heights in the upper

levels: for a cavity width of 0.4m, 34% and 17% of

the total flow was induced at the bottom and top floor

respectively. These floor-to-floor variations were

significantly reduced when PV panels were

integrated at the outer skin of the façade. For this

four-storey structure a cavity width of 0.6-0.8 is

proposed as optimum for the ‘room ventilation’

design.

Punyasompun et al. (2009) performed small

scale experiments for a three-storey building under

the Bangkok climatic conditions and developed a

simplified model that could be used for designing a

‘multi-solar chimney building’. They tested two

possible ways of SC integration; one where each

floor had a separate SC channel (i.e. outlet openings

at every floor), and one where the floors were

connected to a common SC channel, extending from

bottom to top floor (i.e. one outlet opening at the top

of the channel). The latter proved to perform better

and induce higher airflow rates in the floors.

Concluding Remarks

It is evident that fair background knowledge has

been gained so far on the physical mechanisms and

the various parameters affecting the performance of

SC systems. The potential of SC systems to drive

airflow and achieve sufficient passive ventilation in

buildings has been explored in several studies and

was found satisfactory. Most of these studies

however, involved single-room or single-zone

buildings and assumed steady-state conditions, while

most of the outdoor experimental studies (subjected

to dynamic climatic conditions) were performed for

hot and arid climates and specific building types,

typical of the location. Recent studies have

demonstrated the additional complexity and

challenges posed by integrating a SC system with a

multi-storey building, indicating the need for further

research on various related topics.

In practice and for large-scale applications of SC

systems in buildings, it is imperative that the design

be based on performance estimations for dynamic

conditions and for the specific building, climate and

surroundings. Numerical modeling in building

performance simulation (BPS) tools should be

employed when it comes to practical applications:

dynamic thermal and airflow simulations for the SC-

building system can be performed and the feasibility,

potential benefits as well as possible improvements

of the design can be fully explored.

This project studies the feasibility of a SC

integrated in a prototype multi-storey office building

in the Netherlands. Performance simulations will be

performed in the BPS tool ESP-r and results will be

used for sensitivity analysis (SA) and Optimization of

the design. At present only the SA results are

available. The calibration procedure and the SA

results are presented in the following sections.

AIM

The aim of this part of the project is the

investigation of the applicability and optimization of

a solar chimney to enhance natural ventilation in a

multi-storey office building in the Netherlands.

METHODOLOGY

To achieve the aim of the project we envisaged

the research methodology which involves the

following stages (Fig. 3):

1. Calibrating the model of the solar chimney with

the available data from the small-scale

measurements set-up.

2. Performing Sensitivity Analysis (SA) for a SC

design and for the following parameters: length,

cavity width, inside face short-wave absorptivity

and long-wave emissivity, insulation thickness

and thermal mass of the back and side walls,

glazing type and glass percentage of the glazed

wall. For the selected range of variance of the

aforementioned parameters, the Latin Hypercube

sampling method is employed and 200 samples

are generated in total. These are translated to an

equal number of sets of input files for ESP-r and

SA is based on the produced sets of simulation

results. The performance indicators for the SA

are (i) the annual harvested energy (i.e. the air

enthalpic gains in the SC) and (ii) the annual fan

energy savings.

3. Multi-objective optimization of the SC design to

maximize: (i) the air enthalpic gains in the SC

that could be recovered and used for e.g. heating

purposes and (ii) the fan energy savings.

4. Testing the robustness of the final optimized

design using the model of the SC integrated to

the model of the prototype building (e.g.

influence of operating windows, alternate design

for air supply etc).

At the absence of results for stages [4 & 5], the

calibration and SA results are reported in the next

sections. A short description of the case study

building is also included in the last section.

CALIBRATION

Calibration Procedure

The objective of the procedure is to calibrate the

airflow components of the SC model so that pressure

losses, airflow rate and temperatures in the SC

channel are predicted reliably. The calibration is

based on available data from experimental

measurements performed in a small-scale SC test set-

up, constructed in the context of the ‘Earth, Wind &

Fire’ project and located in Mook, Netherlands.

Calibration was performed of the ESP-r model of the

test set-up (i.e. a small-scale SC), which was created

in line with the physical model in terms of geometry

and materials. A presentation of the test set-up and

calibration methodology follows.

The small-scale test set-up.

The SC test set-up (Fig. 4[a]) is 11m high, the

channel is 2m long and 0.25m wide (width-to-height

ratio is 1:44). An aluminum layer with absorptivity of

0.95, emissivity of 0.05 and thickness of 0.5mm is

used for the inside surfaces of the back and side

walls. The walls are well insulated to the outside (U-

value = 0.16 W/m2K) and a double glazing (U-value

= 1.32 W/m2K) is used with a visible light

transmission factor of 0.56 and low-e coating on both

glass panes. The SC channel is connected at its

bottom to a small room (volume is ~20m3) intended

to pre-condition the incoming air up to 20oC. Air

flows naturally into the room through a window

located at its north side (opening is ~1m2) and into

the solar chimney channel through an inlet opening

located close to the floor level (at 0.25m) and which

has the same area and shape as the cross section of

the SC.

An extensive measurements’ schedule was

designed but will not be presented here for brevity

reasons. The data used for the calibration include: (i)

inside surface temperatures for the glazing, back and

side walls at single points and four heights along the

SC channel (0.5/4.0/7.5/9.5m), (ii) air temperatures at

nine points evenly distributed over the channel’s

cross-section and at four heights (0.25/4.0/7.5/11m),

(iii) pressure difference between inlet and outlet of

the channel and (iv) air velocities at two single points

of the cross-section at four heights

(0.25/4.0/7.5/11m). Due to unfavorable weather

conditions and costs involved, measurements were

performed for a limited number of days from

December till April. Measurements from 15/12/09,

08/01/10 and 15/04/10 were used for calibration.

Calibration in ESP-r.

The airflow through the SC is modeled using the

‘general flow conduit’ component, which is assumed

the equivalent of a duct. The required input

parameters for this component are the hydraulic

diameter [m] (equal to the diameter in case of a round

duct), cross-section surface area [m2], length [m],

roughness [m] and local dynamic loss factor [-].

From the above parameters the first three are fixed as

they are determined by the geometry of the SC duct,

while the roughness and local loss factor are the

parameters to be calibrated.

Due to restrictions in the ESP-r program, it was not

possible to impose transient measured boundary

conditions; instead measurement data corresponding

to one-hour timeslots were used and calibration was

performed assuming steady-state conditions for these

hours. Surface temperatures and pressure difference

along the channel are the boundary conditions (input)

to the model. The output data (air temperatures and

mass flow rate) are compared to the corresponding

measured values and the model is calibrated until

reasonable agreement is accomplished. For the air

temperature measurements, the statistical measure of

Figure 4[a,b]. The small –scale experimental set-up

(a) and its model counterpart in ESP-r (b).

R-squared (R2) is used to judge statistical correlation

which can take values from 0 to 1, with 1 being the

absolute agreement between measurements and

model predictions. Since the SC is modeled in ESP-r

with nine thermal zones over the height (see Fig.

4[b]), i.e. model’s input and output data refer to nine

points representing the average conditions in the

zones, the available measured data (at four heights)

were fitted linearly with curves and ‘measured’

values were estimated for the nine model heights, so

that the calculation of the R2 is possible. The R

2 value

of the linear fitting curves was usually above 0.90,

indicating a linear relationship between height and air

temperature in the SC.

It is noted that the measured mass flow value,

derived from velocity measurements at two points of

the SC cross-section (at four different heights), is

assigned an error of 40%, which is the maximum

variation between points along the length and width

of the SC channel, found in measurements performed

specifically to have an idea of the velocity profile

distribution in the channel. Therefore, a valid range

of ±40% around the measured mass flow value is

estimated, within which the predicted value should

fall.

Calibration Results & Discussion

The calibration procedure outlined in the

previous section was performed for ten timeslots. It is

noted that the roughness parameter was found to have

negligible effect in the predictions, therefore it was

fixed at 1mm and calibration consisted of adjusting

the local loss factors of the SC components (Ci). In

all cases a loss factor value between 0.6 and 1.0 was

found to be appropriate (for each timeslot a different

value gave the best results) but since this value is a

fixed input parameter in the model, results are

presented here for a Ci value of 1.0. Agreement

between model predictions and measurements with

respect to air temperatures for every timeslot is

presented in Table 1, expressed with the value of R2,

while agreement of measured and predicted mass

flow rates is presented in Table 2, where predicted

values should fall with the valid range. For brevity

reasons in Table 2 only the number of the timeslot is

included; for more data one can refer to Table 1.

Given the errors introduced by pre-processing of

the measured data (e.g. data fitting with curves and

assumption for the surface temperatures of the walls

at the upper part of the SC) the model’s performance

is considered satisfactory and the calibration

procedure successful. In most cases the R2 value is

above 0.70, whereas lower values are found at times

of low solar radiation intensity values (e.g. timeslot

no.5) or for the timeslots of 15/4/10, where a

significant error is traced back to the trendline of the

surface temperatures. Wind speed is also high during

this day, which could account for the poor

performance of the predictions, since wind is

neglected in the model.

The above are an indication that the model (where the

Ci value is fixed) will not be able to represent mass

flow and air temperatures accurately in some hours of

the day where extreme conditions, in terms of solar

radiation and/or wind speed might come to play a

role in the real system. The complexity of flow

characteristics in a naturally ventilated duct and the

various parameters that affect the flow cannot

possibly be all represented equally well in a model,

which is a simplification of the real system.

The calibrated model was also validated for

whole-day dynamic conditions and results were

found to be satisfactory. Validation results are not

presented here for brevity reasons. Confidence

acquired in the ESP-r model of the SC through

validation allows us to use it for the simulation

predictions required in the next steps.

Table 1. Calibration results: air temperature

agreement between measurements and model

predictions

Table 2. Calibration results: mass flow rate

agreement between measurements and model

predictions.

CASE STUDY

Some information regarding the prototype

building is presented here, since the sensitivity

analysis is performed for a SC design intended to

work for the specific building. The case study is

based on a simplified model of the faculty of

Architecture at the Eindhoven University of

Technology (TU/e) campus, in the Netherlands. The

floor plan of the building is rectangular with

dimensions of 43.2*32.4m (floor surface area is

~1400m2). The 32.4m dimension, corresponding to

the southern and northern façade determines the

maximum possible length of the SC (see next

section). The total building height is 49.5m (9 floors

of 5.5m height each), and ventilation requirements

are estimated at 32.66 m3/s (3.63m

3/s for each floor).

Outdoor air enters the building through a main supply

duct on the northern side (conditioning of the air is

assumed up to 18oC in the supply duct), diverges into

the floors and is consecutively exhausted in the main

exhaust duct on the southern side (exhaust air

temperature is 20oC). The SC channel is connected to

the exhaust duct at ground level; a fan located at the

bottom of the SC will guarantee an airflow of 32.66

m3/s at all times. The pressure drop through the duct

network of the building (excluding the SC channel)

was estimated to be aprox. 40Pa. Dampers are

located at each floor level to balance the system and

ensure that all floors receive the same airflow rate.

The building is assumed to be an office building, thus

the functional hours are from Monday to Friday and

from 7am to 7pm (3132 hours per year).

SENSITIVITY ANALYSIS

Sensitivity analysis is performed for a full-scale

SC of 50m height, intended to be integrated with the

prototype office building previously described. The

parameters considered in the SA are presented in

Table 3, along with the corresponding ranges and

step changes; discrete values and uniform

distributions were considered for each parameter.

Table 3. Parameters and their range of values

considered for SA.

With the term ‘thermal mass’ the thickness of a

concrete layer placed behind the absorbant layer2 of

the back and side walls is considered. Absorptivity

(short-wave) and emissivity (long-wave) are material

2 the term is used here for the surface material of the

opaque walls of the SC, which is usually a low thermal

mass layer with a dark-coloured outer surface for

maximum solar radiation absorption.

properties of the absorbant layer. Three glazing types

are taken into account (single, double and low-e) with

a visible light transmission value of 0.87, 0.76 and

0.79 respectively. Restrictions were imposed so that

the cavity width is less than 25% of the length (to

avoid shading) and that the cross-section area of the

SC results in air velocities not more than 3m/s in the

SC shaft. The parameter ‘glass percentage’ is

considered to account for the reduction of the total

glazed area due structural framing of the glazing

wall. The parameters (independent variables) were

sampled simultaneously using the Latin Hypercube

Sampling method and 200 samples were generated

(for the matrix generation the freeware tool Simlab

was used). The performance indicators considered for

the SA (dependent variables) are (i) the annual

harvested energy in KWh and (ii) the annual fan

energy savings in KWh.

The annual harvested energy is calculated as the sum

of the air enthalpic gains in the SC for all functional

hours:

(1)

Where [m3/s] the volume flow rate through the

SC, [kg/m3] the air density of the incoming air,

[KJ/kgK-1

] the specific heat capacity of air at the

inlet temperature, [oC] the air temperature at the

outlet of the SC and the temperature of the

incoming air, assumed to be 20oC.

The annual fan energy savings are calculated as the

sum for all functional hours of the quantity:

(2)

With the building volume flow rate equal to 32.66

m3/s, =0.65 [-] the fan’s efficiency and the

pressure difference induced by the SC equal to:

(3)

with [kg/m3] the density of the air at the

supply duct (temperature of supply air is assumed

18oC), [kg/m

3] the air density at the SC channel

and [Pa] the pressure losses (local dynamic

and friction losses) of the SC. In case results a

negative quantity the energy savings are set to zero

for that hour, since a negative value would mean that

the fan would have to consume more power to

account for the SC’s high losses. It is reminded that

the fan would need to generate a pressure difference

equal to:

(4)

For the total of 200 samples, simulations were

performed in ESP-r and the two performance

indicators were derived for each sample. The method

of standardized regression coefficients (SRC) was

used to analyze the results, which is a linear

regression method. For every performance indicator

(harvested energy and fan savings) the parameters of

the SA are assigned an SRC value between 0 and 1,

which indicate the sensitivity of the model output to

that parameter. The higher the SRC value, the more

sensitive the model (i.e. the performance indicator) to

that parameter. A positive value of the SRC means

that a positive change of the parameter’s value will

result in a positive change of the performance

indicator. A negative SRC value means that for a

positive change of the parameter a negative change of

the performance indicator is expected. Results of the

SA are presented in the next section.

SA Results

For each performance indicator, the ranking of

the eight parameters, starting from the most

influential, and the SRC values are presented in Table

4. It is noted that the R-square values are 0.875 for

the harvested energy model and 0.871 for the fan

energy savings model, reassuring that the linear

regression method was a good choice; in case these

values were low a non-linear regression analysis

would have been more appropriate.

Table 4. Ranking of the SA parameters.

For the harvested energy the most important

parameters are the length of the SC, the short-wave

absorptivity of the walls, the glazing type (the

positive SRC value here means a double is preferable

to a single glazing and that the low-e would be the

best choice out of the three) and the thermal mass.

For the fan energy savings only the second most

influential parameter is different compared to

harvested energy, and that is the gap width, which

ranks only 7th

in the harvested energy model.

Parameters 5 to 6 are not as important, whereas the

ones ranking 7th

and 8th

could be discarded from the

regression model (the p-value is more than 0.05)3. It

should be noted that the ranking of the insulation

3 The p-value is the probability that the true SRC is zero. If

p-value is higher than 0.05 the variable should be discarded

from the model.

thickness is low because the backwall of the SC is

assumed to be in contact with the wall of the exhaust

duct, where the air temperature is 20oC. The negative

value of the SRC for the thermal mass indicates that

it is preferable to have the lower thermal mass

possible in the walls construction in order to

maximize harvested energy and fan energy savings.

From the results of the 200 simulations a scatter-

plot is produced with the x-axis being the harvested

energy and the y-axis the fan energy savings; each

dot corresponds to one sample. It can be derived from

Figure 5 that the two performance indicators are not

conflicting, i.e. there will be a solution that will

maximise both, as is for example the solution

encircled in the figure. It can also be seen that a large

number of solutions results in zero fan energy

savings (or increased required fan power).

Figure 5. Scatter-plot of the 200 simulation results.

CONCLUSIONS

The applicability of a SC for passive ventilation

of a prototype multi-storey office building in the

Netherlands is explored. Measurement data in a

small-scale SC were used to calibrate and validate the

SC model. Modeling and simulations were performed

in the building simulation tool ESP-r. A case study

office building has been chosen and a sensitivity

analysis (SA) of a SC design intended for this

building was performed. The dimensions of the SC

(length and gap width), the short-wave absorptivity

of the walls, the glazing type and thermal mass were

identified as the most influential parameters with

respect to the annual harvested energy in the SC and

the annual fan energy savings. The next step is the

optimization of the SC design and at the final stage a

robustness analysis of the optimized design against -

among others- active user behavior (e.g. opening of

windows). At the final stage the integrated model of

the SC and the office building will be used.

REFERENCES

Afonso, A. and Oliveira, A. 2000. Solar chimneys:

simulation and experiment. Energy and

Buildings 32: 71-79.

Arce, J., Jimenez, M.J., Guzman, J.D., Heras, M.R.,

Alvarez, G., Xaman, J. 2009. Experimental study

for natural ventilation on a solar chimney.

Renewable Energy 34:2928-2934.

Awbi, H.B. and Gan, G. 1992. Simulation of solar-

induced ventilation. Renewable Energy

Technology and the Environment 4:2016-30.

Bansal, N.K., Mathur, R., Bhandari, M.S. 1994. Solar

chimney for enhanced stack ventilation. Building

and Environment 28:373-377.

Barozzi, G.S., Imbabi, M.S.E., Nobile, E., Sousa,

A.C.M. 1992. Physical and numerical modeling

of a solar chimney-based ventilation system for

buildings. Building and Environment 27:433-

445.

Bassiouny, R. and Koura, N.S.A. 2008. An analytical

and numerical study of solar chimney use for

room natural ventilation. Energy and Buildings

40:865-873.

Borgers, T.R. and Akbari, H. 1979. Laminar flow

within the Trombe wall channel. Solar Energy

22:165-174.

Bouchair, A. 1994. Solar chimney for promoting

cooling ventilation in southern Algeria. Building

Services Engineering Research and Technology

15(2):81-93.

Burek, S.A.M. and Habeb, A. 2007. Air flow and

thermal efficiency characteristics in solar

chimneys and Trombe walls. Energy and

Buildings 39:128-135.

Charvat, P., Jicah, M., Stetina, J. 2004. Solar

chimneys for ventilation and passive cooling.

World Renewable Energy Congress, Denver,

USA.

Chen, Z.D., Bandopadhayay, P., Halldorsson, J.,

Byrjalsen, C., Heiselberg, P., Li, Y. 2003. An

experimental investigation of a solar chimney

model with uniform wall heat flux. Building and

Environment 38:893-906.

Chen, Z.D. and Li, Y. 2001. A numerical study of a

solar chimney with uniform wall heat flux. In:

Proceedings of the Fourth International

Conference on Indoor Air Quality, Ventilation

and Energy Conservation in Buildings, Human,

China, p.1447-54.

Ding, W., Hasemi, Y., Yamada, T. 2005. Natural

ventilation performance of a double-skin façade

with a solar chimney. Energy and Buildings

37:411-418.

Gan, G. and Riffat, S.B. 1998. A numerical study of

solar chimney for natural ventilation of buildings

with heat recovery. Applied Thermal

Engineering 18: 1171 -1187.

Gan, G. 1998. A parametric study of Trombe walls

for passive cooling of buildings. Energy and

Buildings 18:1171-87.

Gan, G. 2006. Simulation of buoyancy-induced flow

in open cavities for natural ventilation. Energy

and Buildings 38:410-420.

Gan, G. 2010. Impact of computational domain on

the prediction of buoyancy-driven ventilation

cooling. Building and Environment 45:1173-

1183.

Harris, D.J., Helwig, N. 2007. Solar chimney and

building ventilation. Applied Energy 84:135-

146.

Hirunlabh, J., Kongduang, W., Namprakai, P.,

Khedari, J. 1999. Study of natural ventilation of

houses by a metallic solar wall under tropical

climate. Renewable Energy 18:109-119.

Ho Lee, K. and Strand, R.K. 2009. Enhancement of

natural ventilation in buildings using a thermal

chimney. Energy and Buildings 41:615-621.

Khedari, J., Boonsri, B., Hirunlabh, J. 2000.

Ventilation impact of a solar chimney on indoor

temperature fluctuation and air change in a

school building. Energy and Buildings 32:89-93.

Khedari, J., Rachapradit, N., Hirunlabh, J. 2003.

Field study of performance of solar chimney

with air-conditioned building. Energy 28:1099-

1114.

Martí-Herrero, J., Heras-Celemin, M.R. 2007.

Dynamic physical model for a solar chimney.

Solar Energy 81:614-622.

Mathur, J., Bansal, N.K., Mathur, N.K., Jain, M.,

Anupma. 2006. Experimental investigations on

solar chimney for room ventilation. Solar Energy

80:927-935.

Nouanégué, H.F. and Bilgen, E. 2009. Heat transfer

by convection, conduction and radiation in solar

chimney systems for ventilation of dwellings.

International Journal of Heat and Fluid Flow

30:150-157.

Ong, K.S. and Chow, C.C. 2003. Performance of a

solar chimney. Solar Energy 74:1-17.

Punyasompun, S., Hirunlabh, J., Khedari, J.

Zeghmati, B. 2009. Investigation on the

application of solar chimney for multi-storey

buildings. Renewable Energy 34:2545-2561.

Rodrigues, A.M., Canha da Piedade, A., Lahellec,

A., Grandpeix, J.Y. 2000. Modeling natural

convection in a heated vertical channel for room

ventilation, Building and Environment 35:455-

469.

SIMLAB. 2009. Version 3.2 Simulation

Environment for Uncertainty and

Sensitivity Analysis, developed by the Joint

Research Centre of the European

Commission.

Spencer, S., Chen, Z.D., Li, Y., Haghighat, F. 2000.

Experimental investigation of a solar chimney

natural ventilation system. In: Proceedings of

Room Vent, Seventh International Conference

on Air Distribution in Rooms, Reading, UK, 9-

12 July, p.813-8.